摘要
【目的】基于深度学习和高光谱成像技术,建立适应果园环境的采前柑橘糖度无损检测模型,为果园柑橘的成熟度判断及智能采摘提供理论参考和技术支持。【方法】采用区域选择法和阈值分割法分别提取柑橘高光谱图像的感兴趣区域(ROI),获得柑橘果实光谱反射的有效区域并计算原始光谱;使用多元散射校正(MSC)、标准正态变换(SNV)、Savitzky-Golay平滑(SG)、小波变换(WT)、一阶导数(FD)、SGMSC组合方法等6种光谱预处理方法分别对原始光谱进行校正,利用竞争性自适应重加权采样算法(CARS)、随机蛙跳算法(RF)、联合区间偏最小二乘法(SIPLS)、相关系数法(CORR)等4种算法提取与柑橘糖度高度相关的特征波段,并与光谱预处理进行组合;在卷积神经网络(CNN)的基础上融合特征增强和残差模块,建立CNN-FA-RB模型,同时构建偏最小二乘回归(PLSR)、支持向量回归(SVR)、前馈神经网络(FFNN)、CNN、AlexNet、ResNet18、SENet18等7种模型进行对比。【结果】以阈值分割法提取的ROI建模效果更优,决定系数(R^(2))较区域选择法平均提高了19.33%。SG+CARS组合为最佳预处理方法,提取得到的25个特征波段覆盖范围较广(399.2~963.0 nm),捕捉到糖分相关的吸收峰(C-H/O-H键振动),建立PLSR模型测试发现,R^(2)为0.709、均方根误差(RMSE)为0.894,即能有效平滑高频噪声,筛选出与糖度强相关的特征波段。CNN-FA-RB模型的糖度检测精度最高且训练速度较快,果园环境下的R^(2)为0.816、RMSE为0.751、平均绝对误差(MAE)为0.576、平均绝对百分比误差(MAPE)为5.616%、训练速度为13.158轮次/s,相对于CNN模型,R^(2)提升了8.66%、RMSE下降了13.97%、MAE下降了15.79%、MAPE下降了14.32%,且训练速度提升了3.71倍。【结论】建立的CNN-FA-RB模型能克服室外环境干扰,进行精准糖度预测。阈值分割+SG+CARS+CNN-FA-RB的方案能有效实现采前柑橘糖度无损检测,为柑橘成熟度判断、机器人采摘、果园监测等场景提供技术支持。
【Objective】Based on deep learning and hyperspectral imaging technology,a non-destructive detection model for pre-harvest citrus sugar content(SC)adapted to orchard environments was established,providing theoretical reference and technical support for citrus maturity assessment and intelligent harvesting in orchards.【Method】The region selection method and the threshold segmentation method were employed to extract the region of interest(ROI)from citrus hyperspectral images respectively,to obtain the effective spectral reflectance area of citrus fruit and calculate the raw spectra.Six spectral preprocessing methods,including multiplicative scatter correction(MSC),standard normal variate(SNV),Savitzky-Golay smoothing(SG),wavelet transform(WT),first derivative(FD),and the SGMSC combination,were applied to correct the raw spectra.Four feature band extraction algorithms,namely competitive adaptive reweighted sampling(CARS),random frog(RF),synergy interval partial least squares(SIPLS),and correlation coefficient(CORR),were used to select feature bands highly correlated with citrus SC,and combined with spectral preprocessing.By integrating feature augmentation and residual blocks into a convolutional neural network(CNN),the CNN-FA-RB model was established.Meanwhile,seven models,including partial least squares regression(PLSR),support vector regression(SVR),feed-forward neural network(FFNN),CNN,AlexNet,ResNet18,and SENet18,were constructed for comparison.【Result】The ROI extracted by the threshold segmentation method yielded superior modeling performance,with the coefficient of determination(R^(2))improved by an average of 19.33%compared with the region selection method.The SG+CARS combination was identified as the optimal preprocessing strategy,extracting 25 feature bands covering a broad range(399.2-963.0 nm)and capturing sugar-related absorption peaks(C-H/O-H bond vibrations).The PLSR model test showed an R^(2)of 0.709 and a root mean square error(RMSE)of 0.894,demonstrating effective smoothing of high-frequency noise and selection of feature bands strongly correlated with SC.CNN-FA-RB model achieved the highest SC detection accuracy with a relatively fast training speed.Under orchard conditions,its R^(2)was 0.816,RMSE was 0.751,mean absolute error(MAE)was 0.576,mean absolute percentage error(MAPE)was 5.616%,and the training speed was 13.158 epoch/s.Compared with the CNN model,R^(2)increased by 8.66%,RMSE decreased by 13.97%,MAE decreased by 15.79%,MAPE decreased by 14.32%,and the training speed improved by 3.71 times.【Conclusion】The constructed CNN-FA-RB model can overcome outdoor environmental interference and achieve accurate SC prediction.The integrated scheme of threshold segmentation+SG+CARS+CNN-FA-RB can effectively realize non-destructive detection of pre-harvest citrus SC,providing technical support for citrus maturity assessment,robotic harvesting,orchard monitoring,and other application scenarios.
作者
陈贺
吴赛赛
曾志康
柴秀娟
何海碧
韦莹莹
蓝玉萍
方辉
李诚瑜
莫小香
CHEN He;WU Sai-sai;ZENG Zhi-kang;CHAI Xiu-juan;HE Hai-bi;WEI Ying-ying;LAN Yu-ping;FANG Hui;LI Cheng-yu;MO Xiao-xiang(Agricultural Science and Technology Information Research Institute,Guangxi Academy of Agricultural Sciences,Nanning,Guangxi 530007,China;Agricultural Information Institute,Chinese Academy of Agricultural Sciences,Beijing 100081,China;Yangshuo Agricultural Technology Extension Center,Guilin,Guangxi 541999,China;School of Artificial Intelligence,Guangxi Minzu University,Nanning,Guangxi 530006,China)
出处
《南方农业学报》
北大核心
2026年第2期299-312,共14页
Journal of Southern Agriculture
基金
广西科技重大专项(桂科AA22036002)
广西农业科学院基本科研业务专项(桂农科2024YP066,桂农科2025ZX12)
南宁市科技重大专项(20222065)。
关键词
柑橘
糖度
高光谱
CNN-FA-RB模型
采前无损检测
citrus
sugar content
hyperspectrum
CNN-FA-RB model
pre-harvest non-destructive detection
